How to add chapters and artwork to a podcast MP3

Finish the edit and loudness processing first. Then export an MP3 and open Episode Prep. Choose that file, up to 150 MB, and the tool will read it on your device. Large recordings work best in a desktop browser.

The check panel reports duration, channels, sample rate and average bitrate. The basic file panel does not listen for mistakes or measure loudness. The separate optional Audio check measures loudness and sample peaks; neither check certifies platform acceptance. Keep your original master and use your ears as part of the final check.

Existing details appear in the form. Set the episode title, podcast name and author. Episode number and year are optional. Save a show preset if you want the podcast name and author to fill missing fields on future episodes in this browser. That preset does not store audio, artwork or your episode title.

Choose a JPEG or PNG cover under 8 MB. This artwork goes inside the MP3. Your show's RSS cover is a separate upload through your host: Apple accepts square JPEG or PNG show covers from 1400 to 3000 pixels per side and requires a solid background without transparency. An image embedded in the MP3 does not update that RSS cover.

Select Speech transcription and, if wanted, Chapter suggestions in the options panel. Each starts off. Speech models are about 80 MB, chapter models about 24 MB, with a shared processing download of about 13 MB. Then press Start in the transcript workspace for English recordings up to one hour. No installer, extension or account is needed; this device does the processing. No recording or transcript is sent to a cloud AI service. Processing speed and memory use depend on your device; a computer is recommended, and the tab needs to remain open.

Save models for later in this browser is optional and off by default. Enable it to reuse model downloads on later visits, or leave it off for temporary model use. Each pack has its own Remove button for saved files. Saved files can reduce downloads, but model metadata may still need an internet connection. Neither option stores your audio or transcript. Website files can still use your browser's ordinary cache.

If you already have a transcript, paste it or import TXT, SRT or VTT instead of downloading the speech model. Plain text works for writing; chapters and moments need real subtitle timestamps. Review the transcript, especially names and unusual terms. Expand Review and edit transcript to correct a passage; its timestamp can seek the original audio preview. Download TXT for the words, SRT or VTT for subtitles, or JSON for timed segments. Timing is approximate. Download before clearing or leaving the page because the transcript is not saved automatically.

The chapter model compares nearby passages to help find topic changes. Suggested titles are phrases extracted from the recording's transcript. Review and edit these drafts, then choose Use these chapters to copy them into the chapter list. If the file already has chapters, the button explicitly replaces that list. Changes to the transcript or chapter spacing need a new suggestions run; they do not silently overwrite your chapter edits.

Audio check uses no AI model. Start it to measure gated integrated loudness in LUFS, sample peaks, stereo balance and long quiet sections. It shows a decoded-audio memory estimate first and flags timestamps to preview. A sample peak is not a true-peak measurement. Quiet sections or high samples are review prompts, not automatic instructions to remove or repair anything.

Moments also needs no extra AI model once a timed transcript exists. Start Find moments, listen from a proposed start and adjust the times. Export a selected excerpt as a 16-bit WAV, or edit a quote card and download a square PNG. The recording remains unchanged. WAV export temporarily decodes audio, so its memory limit may require a shorter source file.

Writing studio is a separate experimental option with about 1.42 GB of models and several GB of working memory. Select only the formats you want: titles, show notes, a short blog, newsletter, a social caption or an X thread. Choose the full transcript or a specific excerpt. Long transcripts are processed in sections and condensed into working notes that may omit details or contain errors. Generation can take many minutes. Check every fact, implication and format, and expect to edit or rewrite drafts before use.

One processing task runs at a time. Cancel stops a worker and preserves completed drafts or reports. Native browser audio decoding may take a moment to finish after cancellation, during which other heavy tools remain paused. Nothing is published or sent to a social account by these tools.

Enter one timestamp and title per line, such as 00:00 Introduction and 03:20 The interview. Use H:MM:SS for recordings over an hour. For more precision, add milliseconds: 03:20.500. Start times must increase, and every chapter must begin before the recording ends.

Use chapter names that describe a moment someone might want to revisit. Chapters are optional, and existing MP3 chapters will populate the field. Episode Prep supports up to 200 chapters and exports one chronological menu. A file with nested chapter menus will be flattened.

Choose Prepare download, then Download MP3. Or use Publishing pack to choose the available files for a ZIP, including the prepared MP3, artwork, transcript, chapters, episode details and writing drafts. A file list shows exactly what the ZIP includes. The exported copy contains your episode details, cover and embedded ID3 chapters. Its audio is copied without re-encoding, so this step does not change the recording's sound or reduce its quality.

Download text gives you a timestamp list to adapt for your episode description. Download JSON gives you a Podcasting 2.0 chapter file for a compatible hosting workflow. Support differs between hosts and players, and a host that processes uploaded audio may remove embedded tags. Check what your host supports and inspect the published result.

Upload the MP3 to your usual host, then complete its episode title, description and artwork fields. Embedded tags are additional file information; they do not replace that publishing form or make changes to your RSS feed.

Listen to the downloaded copy, confirm that the episode is complete, and check the chapters in your target listening apps after publishing. Closing or reloading Episode Prep clears the loaded file. Your downloaded files and any show preset you explicitly saved remain on your device.